Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

From: Torsten Hoefler (htor_at_[hidden])
Date: 2007-10-12 07:00:10


Hello,
> The code was pretty simple. I was trying to send 8MB data from one
> rank to other in a loop(say 1000 iterations). And then i was taking the
> average of time taken and was calculating the bandwidth.
>
> The above logic i tried with both mpirun-with-mca-parameters and without
> any parameters. And to my surprise, the performance was degrading when i
> was trying to manipulate.
That sounds strange. So did you re-use the communication buffers? Did
you try to run some existing benchmarks like Netpipe [1], IMB or
Netgauge [2]?

> Now I have another question in mind. Is it possible to have IB Hardware
> Multicast implementation in OpenMPI? I have gone through the
> issues/challenges for the same, but also read couple of people who have
> successfully done it for Ethernet/Giga-bit Ethernet and IPoIB ofcourse in
> experimental stage. Actually i want to contribute for it in OpenMPI and
> need the help for the same.
As far as I know, there are two groups/people working on this. Andy
Friedley implements a "traditional" ACK based approach (like the one
that the OSU folks published about some time ago) and I implemented a
new idea for extreme scale (see "A practically constant-time MPI
Broadcast Algorithm for large-scale InfiniBand Clusters with
Multicast" [3]). I know that my version is still unstable and has some
problems. But I'm working on this.

Best,
  Torsten

[1]: http://www.scl.ameslab.gov/netpipe/
[2]: http://www.unixer.de/research/netgauge/
[3]: https://www.unixer.de/publications/#hoefler-cac07

-- 
 bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ -----
Computer scientists are the historians of computing.  -- Gordon Bell